The assimilation of satellite-derived sea surface temperatures into a diurnal cycle model

Satellite sea surface temperature (SST) observations available from infrared and microwave radiometers derive a skin and sub-skin temperature measurement from very near the ocean surface. These measurements, particularly those taken during the day under clear calm conditions, are often seen to have a diurnal warming signal. These diurnal SST signals can result in errors and aliasing when observations collected at different times of the day and from a variety of observation sources are merged together to obtain "foundation'' or bulk SST observation products. A similar problem occurs when SST observations are assimilated into ocean models, which typically do not resolve a diurnal cycle. In this article, a novel data assimilation method is developed and implemented that explicitly utilizes diurnal signal information in SST observations. The technique assimilates SST observations taken over the day into a diurnal cycle model by making corrections, within uncertainty bounds, to the surface boundary forcing of the model. In particular, the surface wind speeds and the fractional cloud cover parameter, which are typically poorly known over the oceans, are tuned in the process. This method is shown to improve the estimate of diurnal SST variability, and it has the potential to reduce uncertainties in estimates of foundation or bulk SST. As such, the procedure can be viewed as a dynamic observation operator.